#importing libraries to retrieve data and read csv and to perform EDa tasks
import pandas as pd
import numpy as np
#importing for visualisation
import matplotlib.pyplot as plt
%matplotlib inline
import seaborn as sns
# data preprocessing
from sklearn.preprocessing import StandardScaler
# data splitting
from sklearn.model_selection import train_test_split
from collections import Counter
# data modeling
from sklearn.metrics import confusion_matrix,accuracy_score,roc_curve,classification_report
from sklearn.linear_model import LogisticRegression
from sklearn.naive_bayes import GaussianNB
from sklearn.ensemble import RandomForestClassifier
from sklearn.model_selection import GridSearchCV
from sklearn.tree import DecisionTreeClassifier
from sklearn.neighbors import KNeighborsClassifier
from sklearn.svm import SVC
from xgboost import XGBClassifier

## Read the CSV File Using Pandas read_csv function
data_set=pd.read_csv("C:\\Users\\sumas\\Documents\\Data science\\o8t_test data set\\o8t_testdata.csv")

data_set.columns

Index(['Subject_ID', 'Label', 'Gender', 'Age', 'Education', 'Height', 'Weight',
       'History of cerebrovascular disease', 'History of hypertension',
       'History of diabetes', 'History of coronary heart disease',
       'History of hyperlipidemia', 'History of anemia',
       'History of CO poisoning',
       'History of general anesthesia during surgery',
       'History of abnormal thyroid function',
       'History of traumatic brain injury', 'Family history of dementia',
       'Smoking history', 'Drinking history', 'Unnamed: 21', 'NPI', 'MoCAB',
       'MMSE', 'IADL', 'HAMA', 'HAMD', 'C1 HVLT(immediate memory)',
       'C5 HVLT（delayed recall 5min)', 'C8 HVLT（delayed recall 20min)',
       'C4 logical memory（WMS）', 'C6 Boston Naming Test',
       'C3 articulateness and verbal fluencey-vegetable （BNT）', 'C7-STT_A',
       'C7-STT_B', 'C2 CFT Rey-O（imitation）', 'C9 CFT Rey-O（recall）',
       'HD1 depressive mood', 'HD2 guilty', 'HD3 suicidal',
       'HD7 work & interests', 'HA6 Total Score of Depressive Mood \n',
       'Total score of Depression core factors', 'HA1 Anxiety', 'HA2 Tension',
       'HA3 Fear', 'HA14 Interview perfomance',
       'Total score of Anxiety factors'],
      dtype='object')

data_set.head(5)

	Subject_ID	Label	Gender	Age	Education	Height	Weight	History of cerebrovascular disease	History of hypertension	History of diabetes	...	HD2 guilty	HD3 suicidal	HD7 work & interests	HA6 Total Score of Depressive Mood \n	Total score of Depression core factors	HA1 Anxiety	HA2 Tension	HA3 Fear	HA14 Interview perfomance	Total score of Anxiety factors
0	1	MCI	Male	77	165.0	90.0	9	Yes	Yes	No	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
1	2	Dementia	Male	81	169.0	70.0	0	Yes	Yes	No	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	1.0	0.0	1.0
2	3	Normal	Female	77	155.0	55.0	12	No	Yes	No	...	0.0	0.0	0.0	2.0	2.0	2.0	0.0	0.0	0.0	2.0
3	4	Normal	Male	75	169.0	75.0	9	No	No	No	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
4	5	Normal	Male	68	NaN	NaN	9	No	No	No	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0

5 rows × 48 columns

data_set.dtypes

Subject_ID                                                 int64
Label                                                     object
Gender                                                    object
Age                                                        int64
Education                                                float64
Height                                                   float64
Weight                                                     int64
History of cerebrovascular disease                        object
History of hypertension                                   object
History of diabetes                                       object
History of coronary heart disease                         object
History of hyperlipidemia                                 object
History of anemia                                         object
History of CO poisoning                                   object
History of general anesthesia during surgery              object
History of abnormal thyroid function                      object
History of traumatic brain injury                         object
Family history of dementia                                object
Smoking history                                           object
Drinking history                                          object
Unnamed: 21                                              float64
NPI                                                      float64
MoCAB                                                    float64
MMSE                                                     float64
IADL                                                     float64
HAMA                                                     float64
HAMD                                                     float64
C1 HVLT(immediate memory)                                float64
C5 HVLT（delayed recall 5min)                             float64
C8 HVLT（delayed recall 20min)                            float64
C4 logical memory（WMS）                                   float64
C6 Boston Naming Test                                    float64
C3 articulateness and verbal fluencey-vegetable （BNT）    float64
C7-STT_A                                                  object
C7-STT_B                                                  object
C2 CFT Rey-O（imitation）                                  float64
C9 CFT Rey-O（recall）                                     float64
HD1 depressive mood                                      float64
HD2 guilty                                               float64
HD3 suicidal                                             float64
HD7 work & interests                                     float64
HA6 Total Score of Depressive Mood \n                    float64
Total score of Depression core factors                   float64
HA1 Anxiety                                              float64
HA2 Tension                                              float64
HA3 Fear                                                 float64
HA14 Interview perfomance                                float64
Total score of Anxiety factors                           float64
dtype: object

### summary of given data set ###

data_set.describe()

	Subject_ID	Age	Education	Height	Weight	Unnamed: 21	NPI	MoCAB	MMSE	IADL	...	HD2 guilty	HD3 suicidal	HD7 work & interests	HA6 Total Score of Depressive Mood \n	Total score of Depression core factors	HA1 Anxiety	HA2 Tension	HA3 Fear	HA14 Interview perfomance	Total score of Anxiety factors
count	217.000000	217.000000	215.000000	215.000000	217.000000	0.0	215.000000	215.000000	215.000000	215.000000	...	216.000000	216.000000	216.000000	216.000000	216.000000	215.000000	215.000000	215.000000	215.000000	215.000000
mean	109.000000	71.589862	162.423256	62.200000	10.451613	NaN	2.883721	16.944186	22.627907	17.972093	...	0.277778	0.092593	0.379630	0.648148	1.879630	0.748837	0.530233	0.148837	0.120930	1.548837
std	62.786676	7.582417	7.912631	10.014685	4.500896	NaN	4.836627	7.385680	6.511755	6.044165	...	0.679717	0.348740	0.762097	0.833204	2.657915	1.005678	0.874225	0.479667	0.379718	1.961352
min	1.000000	50.000000	138.000000	41.000000	0.000000	NaN	0.000000	0.000000	0.000000	14.000000	...	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000
25%	55.000000	66.000000	157.000000	55.000000	9.000000	NaN	0.000000	11.000000	20.000000	14.000000	...	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000
50%	109.000000	72.000000	162.000000	60.000000	9.000000	NaN	1.000000	18.000000	25.000000	16.000000	...	0.000000	0.000000	0.000000	0.000000	1.000000	0.000000	0.000000	0.000000	0.000000	1.000000
75%	163.000000	77.000000	169.000000	69.000000	14.000000	NaN	3.000000	23.000000	27.000000	19.500000	...	0.000000	0.000000	0.000000	1.000000	3.000000	1.500000	1.000000	0.000000	0.000000	2.000000
max	217.000000	92.000000	182.000000	92.000000	20.000000	NaN	27.000000	29.000000	30.000000	54.000000	...	3.000000	2.000000	3.000000	3.000000	13.000000	3.000000	3.000000	2.000000	2.000000	8.000000

8 rows × 31 columns

data_set.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 217 entries, 0 to 216
Data columns (total 48 columns):
 #   Column                                                 Non-Null Count  Dtype  
---  ------                                                 --------------  -----  
 0   Subject_ID                                             217 non-null    int64  
 1   Label                                                  212 non-null    object 
 2   Gender                                                 217 non-null    object 
 3   Age                                                    217 non-null    int64  
 4   Education                                              215 non-null    float64
 5   Height                                                 215 non-null    float64
 6   Weight                                                 217 non-null    int64  
 7   History of cerebrovascular disease                     217 non-null    object 
 8   History of hypertension                                217 non-null    object 
 9   History of diabetes                                    217 non-null    object 
 10  History of coronary heart disease                      217 non-null    object 
 11  History of hyperlipidemia                              217 non-null    object 
 12  History of anemia                                      217 non-null    object 
 13  History of CO poisoning                                217 non-null    object 
 14  History of general anesthesia during surgery           217 non-null    object 
 15  History of abnormal thyroid function                   217 non-null    object 
 16  History of traumatic brain injury                      217 non-null    object 
 17  Family history of dementia                             217 non-null    object 
 18  Smoking history                                        217 non-null    object 
 19  Drinking history                                       217 non-null    object 
 20  Unnamed: 21                                            0 non-null      float64
 21  NPI                                                    215 non-null    float64
 22  MoCAB                                                  215 non-null    float64
 23  MMSE                                                   215 non-null    float64
 24  IADL                                                   215 non-null    float64
 25  HAMA                                                   215 non-null    float64
 26  HAMD                                                   216 non-null    float64
 27  C1 HVLT(immediate memory)                              213 non-null    float64
 28  C5 HVLT（delayed recall 5min)                           213 non-null    float64
 29  C8 HVLT（delayed recall 20min)                          213 non-null    float64
 30  C4 logical memory（WMS）                                 213 non-null    float64
 31  C6 Boston Naming Test                                  213 non-null    float64
 32  C3 articulateness and verbal fluencey-vegetable （BNT）  213 non-null    float64
 33  C7-STT_A                                               213 non-null    object 
 34  C7-STT_B                                               213 non-null    object 
 35  C2 CFT Rey-O（imitation）                                213 non-null    float64
 36  C9 CFT Rey-O（recall）                                   213 non-null    float64
 37  HD1 depressive mood                                    216 non-null    float64
 38  HD2 guilty                                             216 non-null    float64
 39  HD3 suicidal                                           216 non-null    float64
 40  HD7 work & interests                                   216 non-null    float64
 41  HA6 Total Score of Depressive Mood 
                   216 non-null    float64
 42  Total score of Depression core factors                 216 non-null    float64
 43  HA1 Anxiety                                            215 non-null    float64
 44  HA2 Tension                                            215 non-null    float64
 45  HA3 Fear                                               215 non-null    float64
 46  HA14 Interview perfomance                              215 non-null    float64
 47  Total score of Anxiety factors                         215 non-null    float64
dtypes: float64(28), int64(3), object(17)
memory usage: 81.5+ KB

#since the dataset contain null values ###
#count total rows in each column which contain null values ####

data_set.isna().sum()

Subject_ID                                                 0
Label                                                      5
Gender                                                     0
Age                                                        0
Education                                                  2
Height                                                     2
Weight                                                     0
History of cerebrovascular disease                         0
History of hypertension                                    0
History of diabetes                                        0
History of coronary heart disease                          0
History of hyperlipidemia                                  0
History of anemia                                          0
History of CO poisoning                                    0
History of general anesthesia during surgery               0
History of abnormal thyroid function                       0
History of traumatic brain injury                          0
Family history of dementia                                 0
Smoking history                                            0
Drinking history                                           0
Unnamed: 21                                              217
NPI                                                        2
MoCAB                                                      2
MMSE                                                       2
IADL                                                       2
HAMA                                                       2
HAMD                                                       1
C1 HVLT(immediate memory)                                  4
C5 HVLT（delayed recall 5min)                               4
C8 HVLT（delayed recall 20min)                              4
C4 logical memory（WMS）                                     4
C6 Boston Naming Test                                      4
C3 articulateness and verbal fluencey-vegetable （BNT）      4
C7-STT_A                                                   4
C7-STT_B                                                   4
C2 CFT Rey-O（imitation）                                    4
C9 CFT Rey-O（recall）                                       4
HD1 depressive mood                                        1
HD2 guilty                                                 1
HD3 suicidal                                               1
HD7 work & interests                                       1
HA6 Total Score of Depressive Mood \n                      1
Total score of Depression core factors                     1
HA1 Anxiety                                                2
HA2 Tension                                                2
HA3 Fear                                                   2
HA14 Interview perfomance                                  2
Total score of Anxiety factors                             2
dtype: int64

#'duplicated()' function in pandas return the duplicate row as True and other as False
#for counting the duplicate elements we sum all the rows
sum(data_set.duplicated())

0

### removing undefined columns ####
data_set.drop(columns=["Unnamed: 21"],inplace=True)

data_set

	Subject_ID	Label	Gender	Age	Education	Height	Weight	History of cerebrovascular disease	History of hypertension	History of diabetes	...	HD2 guilty	HD3 suicidal	HD7 work & interests	HA6 Total Score of Depressive Mood \n	Total score of Depression core factors	HA1 Anxiety	HA2 Tension	HA3 Fear	HA14 Interview perfomance	Total score of Anxiety factors
0	1	MCI	Male	77	165.0	90.0	9	Yes	Yes	No	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
1	2	Dementia	Male	81	169.0	70.0	0	Yes	Yes	No	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	1.0	0.0	1.0
2	3	Normal	Female	77	155.0	55.0	12	No	Yes	No	...	0.0	0.0	0.0	2.0	2.0	2.0	0.0	0.0	0.0	2.0
3	4	Normal	Male	75	169.0	75.0	9	No	No	No	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
4	5	Normal	Male	68	NaN	NaN	9	No	No	No	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
212	213	MCI	Male	70	170.0	85.0	12	Yes	Yes	Yes	...	1.0	0.0	2.0	2.0	8.0	2.0	2.0	0.0	0.0	4.0
213	214	NaN	Male	76	172.0	73.0	12	No	No	No	...	1.0	2.0	2.0	2.0	9.0	2.0	1.0	1.0	1.0	5.0
214	215	NaN	Female	78	158.0	58.0	6	No	No	Yes	...	0.0	0.0	0.0	0.0	1.0	0.0	0.0	0.0	0.0	0.0
215	216	NaN	Male	66	172.0	77.0	12	Yes	Yes	No	...	2.0	0.0	1.0	1.0	5.0	0.0	0.0	0.0	0.0	0.0
216	217	NaN	Female	61	169.0	78.0	12	No	Yes	No	...	0.0	0.0	1.0	2.0	3.0	0.0	0.0	0.0	0.0	0.0

217 rows × 47 columns

data_set["Label"].unique()

array(['MCI', 'Dementia', 'Normal', 'MCI？Dementia', nan, 'MCI，naMCI_MD'],
      dtype=object)

###  dropping dependent variable rows which are Null, not catergorised###
data_set.drop(labels=[185,209],axis=0,inplace=True)
data_set.dropna(subset=["Label"],inplace=True)

data_set.info()

<class 'pandas.core.frame.DataFrame'>
Int64Index: 210 entries, 0 to 212
Data columns (total 47 columns):
 #   Column                                                 Non-Null Count  Dtype  
---  ------                                                 --------------  -----  
 0   Subject_ID                                             210 non-null    int64  
 1   Label                                                  210 non-null    object 
 2   Gender                                                 210 non-null    object 
 3   Age                                                    210 non-null    int64  
 4   Education                                              208 non-null    float64
 5   Height                                                 208 non-null    float64
 6   Weight                                                 210 non-null    int64  
 7   History of cerebrovascular disease                     210 non-null    object 
 8   History of hypertension                                210 non-null    object 
 9   History of diabetes                                    210 non-null    object 
 10  History of coronary heart disease                      210 non-null    object 
 11  History of hyperlipidemia                              210 non-null    object 
 12  History of anemia                                      210 non-null    object 
 13  History of CO poisoning                                210 non-null    object 
 14  History of general anesthesia during surgery           210 non-null    object 
 15  History of abnormal thyroid function                   210 non-null    object 
 16  History of traumatic brain injury                      210 non-null    object 
 17  Family history of dementia                             210 non-null    object 
 18  Smoking history                                        210 non-null    object 
 19  Drinking history                                       210 non-null    object 
 20  NPI                                                    208 non-null    float64
 21  MoCAB                                                  208 non-null    float64
 22  MMSE                                                   208 non-null    float64
 23  IADL                                                   208 non-null    float64
 24  HAMA                                                   208 non-null    float64
 25  HAMD                                                   209 non-null    float64
 26  C1 HVLT(immediate memory)                              206 non-null    float64
 27  C5 HVLT（delayed recall 5min)                           206 non-null    float64
 28  C8 HVLT（delayed recall 20min)                          206 non-null    float64
 29  C4 logical memory（WMS）                                 206 non-null    float64
 30  C6 Boston Naming Test                                  206 non-null    float64
 31  C3 articulateness and verbal fluencey-vegetable （BNT）  206 non-null    float64
 32  C7-STT_A                                               206 non-null    object 
 33  C7-STT_B                                               206 non-null    object 
 34  C2 CFT Rey-O（imitation）                                206 non-null    float64
 35  C9 CFT Rey-O（recall）                                   206 non-null    float64
 36  HD1 depressive mood                                    209 non-null    float64
 37  HD2 guilty                                             209 non-null    float64
 38  HD3 suicidal                                           209 non-null    float64
 39  HD7 work & interests                                   209 non-null    float64
 40  HA6 Total Score of Depressive Mood 
                   209 non-null    float64
 41  Total score of Depression core factors                 209 non-null    float64
 42  HA1 Anxiety                                            208 non-null    float64
 43  HA2 Tension                                            208 non-null    float64
 44  HA3 Fear                                               208 non-null    float64
 45  HA14 Interview perfomance                              208 non-null    float64
 46  Total score of Anxiety factors                         208 non-null    float64
dtypes: float64(27), int64(3), object(17)
memory usage: 78.8+ KB

##converting to categorical data
for columns in ["Label","Gender",'History of cerebrovascular disease', 'History of hypertension',
       'History of diabetes', 'History of coronary heart disease',
       'History of hyperlipidemia', 'History of anemia',
       'History of CO poisoning',
       'History of general anesthesia during surgery',
       'History of abnormal thyroid function',
       'History of traumatic brain injury', 'Family history of dementia',
       'Smoking history', 'Drinking history']:
    if(data_set[columns].dtype == 'object'):
        data_set[columns]= data_set[columns].astype('category')
        data_set[columns] = data_set[columns].cat.codes

data_set.head(5)

	Subject_ID	Label	Gender	Age	Education	Height	Weight	History of cerebrovascular disease	History of hypertension	History of diabetes	...	HD2 guilty	HD3 suicidal	HD7 work & interests	HA6 Total Score of Depressive Mood \n	Total score of Depression core factors	HA1 Anxiety	HA2 Tension	HA3 Fear	HA14 Interview perfomance	Total score of Anxiety factors
0	1	1	1	77	165.0	90.0	9	1	1	0	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
1	2	0	1	81	169.0	70.0	0	1	1	0	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	1.0	0.0	1.0
2	3	2	0	77	155.0	55.0	12	0	1	0	...	0.0	0.0	0.0	2.0	2.0	2.0	0.0	0.0	0.0	2.0
3	4	2	1	75	169.0	75.0	9	0	0	0	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
4	5	2	1	68	NaN	NaN	9	0	0	0	...	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0

5 rows × 47 columns

### deleting columns "Education", "Height", "Weight" as they not important to diagnosis of the given disease ####
data_set.drop(axis=1,columns=["Education", "Height", "Weight"],inplace=True)

data_set.info()

<class 'pandas.core.frame.DataFrame'>
Int64Index: 210 entries, 0 to 212
Data columns (total 44 columns):
 #   Column                                                 Non-Null Count  Dtype  
---  ------                                                 --------------  -----  
 0   Subject_ID                                             210 non-null    int64  
 1   Label                                                  210 non-null    int8   
 2   Gender                                                 210 non-null    int8   
 3   Age                                                    210 non-null    int64  
 4   History of cerebrovascular disease                     210 non-null    int8   
 5   History of hypertension                                210 non-null    int8   
 6   History of diabetes                                    210 non-null    int8   
 7   History of coronary heart disease                      210 non-null    int8   
 8   History of hyperlipidemia                              210 non-null    int8   
 9   History of anemia                                      210 non-null    int8   
 10  History of CO poisoning                                210 non-null    int8   
 11  History of general anesthesia during surgery           210 non-null    int8   
 12  History of abnormal thyroid function                   210 non-null    int8   
 13  History of traumatic brain injury                      210 non-null    int8   
 14  Family history of dementia                             210 non-null    int8   
 15  Smoking history                                        210 non-null    int8   
 16  Drinking history                                       210 non-null    int8   
 17  NPI                                                    208 non-null    float64
 18  MoCAB                                                  208 non-null    float64
 19  MMSE                                                   208 non-null    float64
 20  IADL                                                   208 non-null    float64
 21  HAMA                                                   208 non-null    float64
 22  HAMD                                                   209 non-null    float64
 23  C1 HVLT(immediate memory)                              206 non-null    float64
 24  C5 HVLT（delayed recall 5min)                           206 non-null    float64
 25  C8 HVLT（delayed recall 20min)                          206 non-null    float64
 26  C4 logical memory（WMS）                                 206 non-null    float64
 27  C6 Boston Naming Test                                  206 non-null    float64
 28  C3 articulateness and verbal fluencey-vegetable （BNT）  206 non-null    float64
 29  C7-STT_A                                               206 non-null    object 
 30  C7-STT_B                                               206 non-null    object 
 31  C2 CFT Rey-O（imitation）                                206 non-null    float64
 32  C9 CFT Rey-O（recall）                                   206 non-null    float64
 33  HD1 depressive mood                                    209 non-null    float64
 34  HD2 guilty                                             209 non-null    float64
 35  HD3 suicidal                                           209 non-null    float64
 36  HD7 work & interests                                   209 non-null    float64
 37  HA6 Total Score of Depressive Mood 
                   209 non-null    float64
 38  Total score of Depression core factors                 209 non-null    float64
 39  HA1 Anxiety                                            208 non-null    float64
 40  HA2 Tension                                            208 non-null    float64
 41  HA3 Fear                                               208 non-null    float64
 42  HA14 Interview perfomance                              208 non-null    float64
 43  Total score of Anxiety factors                         208 non-null    float64
dtypes: float64(25), int64(2), int8(15), object(2)
memory usage: 52.3+ KB

## retaining only total depression core factors scores and Total score of Anxiety factors, deleting indiviual score columns ####
data_set.drop(axis = 1,columns=["HD1 depressive mood", "HD2 guilty", 'HD3 suicidal','HD7 work & interests', 'HA6 Total Score of Depressive Mood \n'], inplace = True)
data_set.drop( axis = 1,columns=['HA1 Anxiety', 'HA2 Tension','HA3 Fear', 'HA14 Interview perfomance'], inplace = True)

## Handelling missing values ###
data_set.fillna(0,inplace=True)

## replacing "unable to compelete" cells with 
data_set["C7-STT_A"].replace("Unable to complete",0,inplace=True)
data_set["C7-STT_B"].replace("Unable to complete",0,inplace=True)

##converting dtype
for columns in ["C7-STT_A","C7-STT_B"]:data_set[columns]= data_set[columns].astype('float64')

Data analysis and visualisation

from pandas_profiling import ProfileReport
profile = ProfileReport(data_set)
profile

plt.figure(figsize = (40,20))
sns.heatmap(data_set.corr(),annot=True,linewidths=1)

C:\Users\sumas\anaconda3\lib\site-packages\matplotlib\backends\backend_agg.py:238: RuntimeWarning: Glyph 65288 missing from current font.
  font.set_text(s, 0.0, flags=flags)
C:\Users\sumas\anaconda3\lib\site-packages\matplotlib\backends\backend_agg.py:238: RuntimeWarning: Glyph 65289 missing from current font.
  font.set_text(s, 0.0, flags=flags)
C:\Users\sumas\anaconda3\lib\site-packages\matplotlib\backends\backend_agg.py:201: RuntimeWarning: Glyph 65288 missing from current font.
  font.set_text(s, 0, flags=flags)
C:\Users\sumas\anaconda3\lib\site-packages\matplotlib\backends\backend_agg.py:201: RuntimeWarning: Glyph 65289 missing from current font.
  font.set_text(s, 0, flags=flags)

<AxesSubplot:>

sns.countplot(data_set['Label'])

C:\Users\sumas\anaconda3\lib\site-packages\seaborn\_decorators.py:36: FutureWarning: Pass the following variable as a keyword arg: x. From version 0.12, the only valid positional argument will be `data`, and passing other arguments without an explicit keyword will result in an error or misinterpretation.
  warnings.warn(

<AxesSubplot:xlabel='Label', ylabel='count'>

plt.style.use('fivethirtyeight')
plt.figure(figsize=(8,5))
sns.countplot(x=data_set.Label,hue=data_set.Gender)
plt.legend(labels=['Female', 'Male'])

<matplotlib.legend.Legend at 0x20f94c3a580>

data_set.groupby('Gender')['Label'].value_counts()

Gender  Label
0       1        49
        2        39
        0        26
1       2        44
        1        38
        0        14
Name: Label, dtype: int64

data_set.groupby('Label')['Age'].mean()

Label
0    74.775000
1    71.988506
2    69.481928
Name: Age, dtype: float64

fg = sns.FacetGrid(col = 'Label',data=data_set)
fg.map(plt.hist,'Age')

<seaborn.axisgrid.FacetGrid at 0x20fc88c4c10>

Model preparation

##splitting the dataset into the training set and the Test set
y = data_set["Label"]
X = data_set.drop('Label',axis=1)
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.30, random_state = 1)
scaler = StandardScaler()
X_train = scaler.fit_transform(X_train)
X_test = scaler.transform(X_test)

Before applying algorithm we should check whether the data is equally splitted or not, because if data is not splitted equally it will cause for data imbalacing problem

print(y_test.unique())

len(y_train)

[0 2 1]

147

len(y_test)

63

Counter(y_train)

Counter({1: 63, 0: 29, 2: 55})

Counter(y_test)

Counter({0: 11, 2: 28, 1: 24})

Machine learning models

Naive Bayes, Random Forest Classifier, K-Nearest Neighbour, Decision Tree, Support Vector Machine, Multinominal Logistic Regression, Extreme Gradient Boost

### Training the Naive Bayes model and Evaluation
m1 = 'Naive Bayes'
nb = GaussianNB()
nb.fit(X_train,y_train)
nbpred = nb.predict(X_test)
nb_conf_matrix = confusion_matrix(y_test, nbpred)
nb_acc_score = accuracy_score(y_test, nbpred)
print("confusion matrix")
print(nb_conf_matrix)
print("\n")
print("Accuracy of Naive Bayes model:",nb_acc_score*100,'\n')
print(classification_report(y_test,nbpred))

confusion matrix
[[11  0  0]
 [ 8 14  2]
 [ 0  4 24]]


Accuracy of Naive Bayes model: 77.77777777777779 

              precision    recall  f1-score   support

           0       0.58      1.00      0.73        11
           1       0.78      0.58      0.67        24
           2       0.92      0.86      0.89        28

    accuracy                           0.78        63
   macro avg       0.76      0.81      0.76        63
weighted avg       0.81      0.78      0.78        63

While training the Naive Bayes model in the dataset, we achieved an accuracy rate of 78%, which is more than the baseline accuracy. For classifying class “0” (i.e.Dementia) the algorithm achieved a recall value of 1 and an F1 score of 0.73 and for class "1"(MCI) the recall value of 0.6 and F1 score of 0.67

### Training the Random Forest model and Evaluation
m2 = 'Random Forest Classfier'
param_grid={"n_estimators":[10,20,40,60,80,100,120,140,160,180,200]}
rf = GridSearchCV(RandomForestClassifier(),param_grid,refit=True,verbose=0)
rf.fit(X_train,y_train)
rf_predicted = rf.predict(X_test)
rf_conf_matrix = confusion_matrix(y_test, rf_predicted)
rf_acc_score = accuracy_score(y_test, rf_predicted)
print("confusion matrix")
print(rf_conf_matrix)
print("\n")
print("Accuracy of Random Forest:",rf_acc_score*100,'\n')
print(classification_report(y_test,rf_predicted))

confusion matrix
[[ 7  4  0]
 [ 0 23  1]
 [ 0  2 26]]


Accuracy of Random Forest: 88.88888888888889 

              precision    recall  f1-score   support

           0       1.00      0.64      0.78        11
           1       0.79      0.96      0.87        24
           2       0.96      0.93      0.95        28

    accuracy                           0.89        63
   macro avg       0.92      0.84      0.86        63
weighted avg       0.90      0.89      0.89        63

The Random Forest model attained an accuracy rate of 89%, which is more than the Naive Bayes model. Moreover, the precision for identifying Dementia is 1 with an F1 score of 0.78.

### Training the K-NN model and Evaluation
m3 = 'K-NeighborsClassifier'
knn = KNeighborsClassifier(n_neighbors=7)
knn.fit(X_train, y_train)
knn_predicted = knn.predict(X_test)
knn_conf_matrix = confusion_matrix(y_test, knn_predicted)
knn_acc_score = accuracy_score(y_test, knn_predicted)
print("confusion matrix")
print(knn_conf_matrix)
print("\n")
print("Accuracy of K-NeighborsClassifier:",knn_acc_score*100,'\n')
print(classification_report(y_test,knn_predicted))

confusion matrix
[[ 7  4  0]
 [ 0 21  3]
 [ 0  4 24]]


Accuracy of K-NeighborsClassifier: 82.53968253968253 

              precision    recall  f1-score   support

           0       1.00      0.64      0.78        11
           1       0.72      0.88      0.79        24
           2       0.89      0.86      0.87        28

    accuracy                           0.83        63
   macro avg       0.87      0.79      0.81        63
weighted avg       0.85      0.83      0.83        63

The KNN model achieved an accuracy rate of 83%, which is slightly less than the Random Forest model.The model has precision of 1 and f1-scores of 0.78 for identifying the class “0” (Dementia)

### Training the DecisionTree model and Evaluation
m4 = 'DecisionTreeClassifier'
dt = DecisionTreeClassifier(criterion = 'entropy',random_state=0,max_depth = 15)
dt.fit(X_train, y_train)
dt_predicted = dt.predict(X_test)
dt_conf_matrix = confusion_matrix(y_test, dt_predicted)
dt_acc_score = accuracy_score(y_test, dt_predicted)
print("confusion matrix")
print(dt_conf_matrix)
print("\n")
print("Accuracy of DecisionTreeClassifier:",dt_acc_score*100,'\n')
print(classification_report(y_test,dt_predicted))

confusion matrix
[[ 7  4  0]
 [ 1 22  1]
 [ 0  5 23]]


Accuracy of DecisionTreeClassifier: 82.53968253968253 

              precision    recall  f1-score   support

           0       0.88      0.64      0.74        11
           1       0.71      0.92      0.80        24
           2       0.96      0.82      0.88        28

    accuracy                           0.83        63
   macro avg       0.85      0.79      0.81        63
weighted avg       0.85      0.83      0.83        63

The DecisionTree model achieved an accuracy rate of 83%, which is slightly less than the Random Forest model.

### Training the SVM model and Evaluation
m5 = 'Support Vector Classifier'
param_grid = {'C': [0.1,1, 10, 100, 1000], 'gamma': [1,0.1,0.01,0.001,0.0001], 'kernel': ['rbf']} 
svc = GridSearchCV(SVC(),param_grid,refit=True,verbose=0)
svc.fit(X_train, y_train)
svc_predicted = svc.predict(X_test)
svc_conf_matrix = confusion_matrix(y_test, svc_predicted)
svc_acc_score = accuracy_score(y_test, svc_predicted)
print("confusion matrix")
print(svc_conf_matrix)
print("\n")
print("Accuracy of Support Vector Classifier:",svc_acc_score*100,'\n')
print(classification_report(y_test,svc_predicted))

confusion matrix
[[ 6  5  0]
 [ 0 22  2]
 [ 0  2 26]]


Accuracy of Support Vector Classifier: 85.71428571428571 

              precision    recall  f1-score   support

           0       1.00      0.55      0.71        11
           1       0.76      0.92      0.83        24
           2       0.93      0.93      0.93        28

    accuracy                           0.86        63
   macro avg       0.90      0.80      0.82        63
weighted avg       0.88      0.86      0.85        63

SVM shows lower accuracy than random forest model, which is 86%.

### Training the Multinomial logistic regression model and Evaluation
m6="multinomial logistic regression"
mlr= LogisticRegression(multi_class='multinomial', solver='lbfgs')
mlr.fit(X_train, y_train)
mlr_predicted = mlr.predict(X_test)
mlr_conf_matrix = confusion_matrix(y_test, mlr_predicted)
mlr_acc_score = accuracy_score(y_test, mlr_predicted)
print("confusion matrix")
print(mlr_conf_matrix)
print("\n")
print("Accuracy of multinomial logistic regression:",mlr_acc_score*100,'\n')
print(classification_report(y_test,mlr_predicted))

confusion matrix
[[ 6  5  0]
 [ 0 22  2]
 [ 0  5 23]]


Accuracy of multinomial logistic regression: 80.95238095238095 

              precision    recall  f1-score   support

           0       1.00      0.55      0.71        11
           1       0.69      0.92      0.79        24
           2       0.92      0.82      0.87        28

    accuracy                           0.81        63
   macro avg       0.87      0.76      0.79        63
weighted avg       0.85      0.81      0.81        63

### Training the Extreme Gradient Boost model and Evaluation
m7 = 'Extreme Gradient Boost'
xgb = XGBClassifier(learning_rate=0.01, n_estimators=25, max_depth=15,gamma=0.6, subsample=0.52,colsample_bytree=0.6,seed=27, 
                    reg_lambda=2, booster='dart', colsample_bylevel=0.6, colsample_bynode=0.5,use_label_encoder=False,eval_metric='mlogloss')
xgb.fit(X_train, y_train)
xgb_predicted = xgb.predict(X_test)
xgb_conf_matrix = confusion_matrix(y_test, xgb_predicted)
xgb_acc_score = accuracy_score(y_test, xgb_predicted)
print("confusion matrix")
print(xgb_conf_matrix)
print("\n")
print("Accuracy of Extreme Gradient Boost:",xgb_acc_score*100,'\n')
print(classification_report(y_test,xgb_predicted))

confusion matrix
[[ 6  5  0]
 [ 0 22  2]
 [ 0  2 26]]


Accuracy of Extreme Gradient Boost: 85.71428571428571 

              precision    recall  f1-score   support

           0       1.00      0.55      0.71        11
           1       0.76      0.92      0.83        24
           2       0.93      0.93      0.93        28

    accuracy                           0.86        63
   macro avg       0.90      0.80      0.82        63
weighted avg       0.88      0.86      0.85        63

Model Evaluation

model_ev = pd.DataFrame({'Model': ['Naive Bayes','Random Forest','K-Nearest Neighbour','Decision Tree','Support Vector Machine','multinomial logistic regression','Extreme Gradient Boost'], 'Accuracy': [nb_acc_score*100,rf_acc_score*100,knn_acc_score*100,dt_acc_score*100,svc_acc_score*100,mlr_acc_score*100,xgb_acc_score*100]})
model_ev

	Model	Accuracy
0	Naive Bayes	77.777778
1	Random Forest	88.888889
2	K-Nearest Neighbour	82.539683
3	Decision Tree	82.539683
4	Support Vector Machine	85.714286
5	multinomial logistic regression	80.952381
6	Extreme Gradient Boost	85.714286

colors = ['green','blue','gold','silver','yellow','red','black']
plt.figure(figsize=(25,5))
plt.title("barplot Represent Accuracy of different models")
plt.xlabel("Accuracy %")
plt.ylabel("Algorithms")
plt.bar(model_ev['Model'],model_ev['Accuracy'],color = colors)
plt.show()

Conclusion Random Forest gives the best Accuracy compared to other models.